AI Photo Editing for TikTok: A Complete Guide
What survives vertical video compression, why photo edits behave differently inside a clip, and the three changes that actually help.
Most advice about photo editing for TikTok is ordinary photo editing advice with the word TikTok added. The platform imposes a specific constraint that changes which edits are worth making, and it is not the one people expect.
The constraint: your photo becomes video
This is the thing that explains everything else.
A still used in a TikTok is not served as an image. It is encoded into a video stream, and video compression is far more aggressive than image compression — it is optimised for motion, and a static frame gets compressed hard.
What video compression destroys first:
- Fine high-contrast detail. Sharpening turns into visible artifacts.
- Saturated areas. Vivid reds and blues band and block up.
- Subtle gradients. Skies and soft backgrounds posterise.
- Small text. Becomes mushy at typical bitrates.
Which is a near-perfect list of what heavy editing adds. The edits that look most impressive on your phone are the ones the platform is most efficient at removing.
The three edits that survive
What holds up is structural change — editing that alters what is in the frame rather than how the pixels are adjusted.
1. Remove the distraction
A cluttered frame at feed speed reads as noise. Removing the thing pulling the eye away from your subject is the highest-value edit and it survives compression completely, because you have changed content rather than added detail. The technique is in removing unwanted objects.
2. Replace or simplify the background
Increases subject separation, which is what still reads after compression. Also the safest edit available, because your subject is untouched. See background replacement.
3. Fix the light
Lifting a face out of shadow or correcting an indoor colour cast makes a frame legible at a glance. This is tonal work rather than detail work, and tonal contrast survives compression far better than micro-contrast does.
What to stop doing
Sharpening. It is compressed away at best and turns to artifacts at worst.
Pushing saturation. The first thing to band.
Fine texture work. Invisible at 1080 in a video stream.
Editing inside the app. By then the platform has already resized your image. You are adjusting a degraded copy and then compressing it again.
Preparing the file
- 1080 by 1920, prepared at that size rather than cropped in-app
- Edit at full resolution, export once. Every additional encode costs quality
- Check at actual size before uploading — not on a monitor, on a phone
- Avoid tool watermarks. Beyond looking templated, platforms have historically deprioritised content carrying other platforms' marks
Motion beats a better still
Worth saying because it changes priorities.
A static frame in a feed reads as an advertisement and gets scrolled. A few seconds of movement — a slow push, a still animated into a short clip — holds attention through the first second, which is the only part most posts get.
That means: past a modest quality bar, adding motion to an adequate photo beats perfecting a static one. The practical side of producing those clips is in preparing for AI video, and what makes a photo animate well is in how photo to video works.
Why in-app filters behave differently here
A detail that catches people out when they move between platforms.
In-app filters are applied to the version the platform already holds, after its own resize, and then baked into the video encode. So a filter that looks identical in two apps can produce noticeably different results, because each applied it at a different point in a different pipeline.
The practical consequence: do your structural edits outside, at full resolution, and use in-app adjustment only for small things you want to preview against the rest of the clip. Doing the heavy work in-app means the heavy work happens to a degraded copy.
It also means comparing your output against someone else's is not a fair comparison unless you know where in their pipeline the edit happened. A clip that looks cleaner may simply have been prepared outside the app rather than edited more skilfully.
The half-second test, again
Same test as for any feed: shrink the frame to roughly the size it will appear, glance at it, and ask whether there is one obvious subject and whether it separates from its background.
If it does not pass, no amount of adjustment will help — that is a composition problem. And composition is the one thing compression cannot take away.
A workable sequence
- Choose a photo with one clear subject
- Remove distractions and simplify the background
- Fix the light, lightly
- Export at 1080 by 1920, once
- Add motion if the post warrants it
- Check on a phone before posting
The equivalent guidance for still-image feeds, where the constraints differ, is in making Instagram photos pop.
Four of Kitana's tools run in the browser, including background replacement; the thirteen photo tools and Photo to Video are in the apps.
Frequently asked questions
- Why do my edited photos look worse on TikTok than on my phone?
- Because a still inside a TikTok is not displayed as an image — it is encoded as video. Video compression is far more aggressive than image compression, and it targets exactly what heavy editing adds: fine high-contrast detail and saturated areas. A restrained edit survives it; an aggressive one falls apart.
- What resolution should photos be for TikTok?
- 1080 by 1920 for a full-screen vertical still. Anything larger is downscaled; anything smaller is upscaled by the platform, which looks worse than doing it yourself. Prepare the still at the final frame size rather than letting the app crop it.
- Does adding motion to a still help?
- It helps retention in the first second, which is the only part most posts get. A slow push or a photo animated into a few seconds of movement stops the scroll better than a static frame, which the eye reads as an advertisement.
- Should I edit before or after adding the photo to a video?
- Before, at full resolution, outside the app. In-app editing happens after the platform has already resized your image, so you are adjusting a degraded version and then compressing it again.
- What edits hold up best?
- Structural ones: removing a distraction, replacing a cluttered background, fixing the light. They change what the frame contains, which survives compression. Colour and sharpening adjustments are the first things it destroys.
- Is a watermark from an editing app a problem?
- Yes, beyond the obvious. Platforms have historically deprioritised content carrying other platforms' or tools' watermarks, and visually a watermark marks the clip as templated. Check whether your tool watermarks free output before you build a workflow on it.
Ready to put this into practice?
Create with Kitana using the tool that fits this guide.
Prepare a photoReady to try it yourself?
Download Kitana and create your first AI photo in under a minute.